kudu to csv to hive to kudu 1 建表kudu

create table default.kudu_table( id string not null, ts bigint, count bigint, primary key(id) ) PARTITION BY HASH (id) PARTITIONS 6 STORED AS KUDU;

2 插入数据

insert into default.kudu_table values (“1”, 11, 11), (“2”, 22, 22)

3 导出到linux

impala-shell -q ‘select * from default.kudu_table’ -B --output_delimiter="*" -o /data/tmp/test.csv

4 建表hive，临时表

create table default.kudu_table2( id string, ts bigint, count bigint)row format delimited fields terminated by ‘*’;

处理null值问题

alter table default.kudu_table2 SET SERDEPROPERTIES(‘serialization.null.format’ = ‘NULL’);

5 csv导入hive

load data local inpath ‘/data/tmp/test.csv’ into table default.kudu_table2;

6 建表kudu，最终表

create table default.kudu_table3( id string not null, ts bigint, count bigint, primary key(id) ) PARTITION BY HASH (id) PARTITIONS 6 STORED AS KUDU;

7 hive to kudu，刷新impala数据

insert into default.kudu_table3 select * from default.kudu_table2;

---

1.本站遵循行业规范，任何转载的稿件都会明确标注作者和来源；2.本站的原创文章，会注明原创字样,如未注明都非原创，如有侵权请联系删除！；3.作者投稿可能会经我们编辑修改或补充；4.本站不提供任何储存功能只提供收集或者投稿人的网盘链接。